Camelia, the Perl 6 bug

IRC log for #padre, 2010-01-23

| Channels | #padre index | Today | | Search | Google Search | Plain-Text | summary

All times shown according to UTC.

Time Nick Message
00:03 patspam joined #padre
00:32 robn joined #padre
00:58 patspam joined #padre
01:04 CSJewell2 joined #padre
01:18 Ryan52 jq: pong
01:26 submersible joined #padre
01:26 Hyppolit svn: r10238 | submersible_toaster++ | http://padre.perlide.org/trac/changeset/10238
01:26 Hyppolit add geometry and deps on Graph, add Resources (currently disabled) , blatently stolen from padre directory/project browser.
01:26 Hyppolit trunk/Padre-Plugin-Swarm/ trunk/Padre-Plugin-Swarm/lib/Padre/Plugin/ trunk/Padre-Plugin-Swarm/lib/Padre/Plugin/Swarm/ trunk/Padre-Plugin-Swarm/l​ib/Padre/Plugin/Swarm/Wx/ trunk/Padre-Plugin-Swarm/lib/Padre/Swarm/
01:27 CSJewell2 joined #padre
01:32 Alias_ joined #padre
01:38 CSJewell joined #padre
03:19 Alias__ joined #padre
03:48 submersible hi folks
03:52 [1]awnstudio joined #padre
04:15 Hyppolit #828: editor_disable plugin hook missing? (new defect) [ http://padre.perlide.org/trac/ticket/828 ]
04:45 stemd joined #padre
05:18 waxhead joined #padre
05:54 stemd joined #padre
06:06 danlucraft joined #padre
06:16 azawawi joined #padre
06:16 azawawi good morning
06:16 Alias__ heya
06:17 Alias__ azawawi, I can fix the problem with always forcing the splash screen
06:17 Alias__ Hacking on it now
06:17 azawawi i saw some of your changes on Padre::Desktop
06:17 azawawi fixing TODOs. ... :)
06:17 Alias__ Any chance you can take a look at that problem of auto-substituting a false STDOUT and STDERR at script startup if it's being called from padre.exe?
06:18 azawawi i had a really sad week
06:18 Alias__ I don't really like the temporary hack I had to put into ORLite::Migrate
06:18 azawawi my grandmother died and then my uncle...
06:18 Alias__ sad as in lame, or sad as in funeral
06:18 Alias__ ouch :(
06:18 * Alias__ has both grandmothers in the latter stages of Alzeimers
06:18 azawawi i had no time to do anything...
06:18 Alias__ It's totally fine
06:19 Alias__ That's why we give everyone commit so much
06:19 Alias__ So people can have real lives
06:20 * azawawi updates his svn...
06:20 azawawi mibbit turned on websocket support for chrome
06:21 azawawi but so far it is really shaky and flimsy...
06:21 azawawi breaking every week
06:29 stemdA joined #padre
06:39 submersible Alias: do you recall if plugin 'editor_disable' ever worked?
06:39 submersible #828
06:39 Hyppolit #828: editor_disable plugin hook missing? (new defect) [ http://padre.perlide.org/trac/ticket/828 ]
06:46 Haarg joined #padre
06:59 Alias__ submersible, I wasn't involved in the editor-hooking, it dates from WAY back when people were experimenting with Vi support
06:59 Alias__ About 20-30 releases ago
06:59 Alias__ I don't think it really ever got much beyond the shitty first implementation
07:09 Hyppolit svn: r10239 | azawawi++ | http://padre.perlide.org/trac/changeset/10239
07:09 Hyppolit Removed hardcoded strawberry path from --desktop
07:09 Hyppolit trunk/Padre/lib/Padre/
07:12 submersible Alias: it seems the mechanism for _ever_ running 'editor_disable' has since been lost. ..
07:13 submersible what to you think... trap the CLOSE of the editor? and issue the editor_disable that way?
07:14 * submersible shudders at the TreeCtrl gymnastics
07:16 Alias__ The "CLOSE" ?
07:17 Alias__ You need to do it somewhere between when all user-interactive actions are finished, but before the widget Destroy
07:19 submersible oh yeah.. otherwise u get an event w/ a dead editor..
07:19 submersible crap
07:20 Alias__ And probably after any saving of the file
07:22 Alias__ Put a debug trap on Padre::Wx::Main::on_close
07:22 Alias__ And then walk through the close process and see if you can find the obvious place
07:36 cognominal joined #padre
07:39 Astemd joined #padre
08:00 Astemd joined #padre
08:06 stemdA joined #padre
08:14 pece joined #padre
08:19 Astemd joined #padre
08:29 azawawi Alias__: ping
08:29 Alias__ pong
08:29 azawawi Alias__: is writing to registry with Win32::TieRegistry working under vista?
08:29 Alias__ Do you have a reason to suspect it doesn't?
08:30 azawawi Alias__: UAC
08:30 Alias__ Have you actually SEEN it not working?
08:30 azawawi i've been trying to write a key... and so far nothing
08:30 azawawi for Padre::Desktop
08:30 azawawi let me try the same code over an xp machine to verify
08:30 Alias__ Write a .t test
08:31 Alias__ Then turn off UAC on Vista
08:31 Alias__ There's a toggle for it in the Security Centre thing
08:31 azawawi a restart lol
08:31 Alias__ Yup
08:31 Alias__ See you in a bit :)
08:32 azawawi cya
08:38 azawawi joined #padre
08:38 azawawi uac...
08:39 azawawi Alias__: uac is the culprit
08:39 Alias__ darn
08:39 azawawi how can we ask for UAC permission like regedit?
08:39 Alias__ Not sure
08:39 azawawi maybe something in the assembly...
08:40 Alias__ Maybe we Win32::API
08:44 Hyppolit svn: r10240 | azawawi++ | http://padre.perlide.org/trac/changeset/10240
08:44 Hyppolit Converted the registry code to Win32::TieRegistry for --desktop
08:44 Hyppolit trunk/Padre/lib/Padre/
08:53 azawawi alias: http://social.msdn.microsoft.com/for​ums/en-US/windowssecurity/thread/6d8​70287-e364-4151-851d-70605c0e6a6d/
08:53 azawawi Alias: we need to manipulate the manifest... like in wperl.exe.manifest
08:53 * azawawi reboot to enable UAC
08:59 azawawi joined #padre
09:25 Hyppolit svn: r10241 | azawawi++ | http://padre.perlide.org/trac/changeset/10241
09:25 Hyppolit Removing empty QuickFixProvider (AdamK++ moved them to Padre::Document::Perl::QuickFix)
09:25 Hyppolit trunk/Padre/lib/Padre/
09:26 pece2 joined #padre
09:32 Sewi joined #padre
09:32 Sewi gm
09:33 azawawi hi Sewi
09:33 Sewi Hey azawawi, nice to see you again
09:34 azawawi thanks
09:35 azawawi any idea how to elevate padre's execution level (for UAC) at runtime?
09:36 Sewi I don't think that I understand what you mean.
09:37 azawawi remember the user account control dialog that pops up on vista?
09:38 Sewi No, never used it on Vista (never used Vista), used it on XP and W7
09:39 azawawi i need something like sudoing a certain action (e.g. padre --desktop)
09:41 Sewi This might help: http://help.lockergnome.com/vista/Command-Line-​cmd-exe-administrator-access--ftopict1394.html
09:42 Sewi But is the UAC dialog so bad while putting the icon on the desktop? It's something you do once...
09:43 azawawi yeah
09:43 azawawi i need something like the "Show all processes" button on task manager
09:43 Sewi ok
09:43 azawawi i need it to elevate only when --desktop is invoked...
09:48 azawawi Sewi: we need also to drop the calling of wperl.exe and re-use code that someone created to embed perl...
09:49 Hyppolit #829: Switch if-Syntax (new enhancement) [ http://padre.perlide.org/trac/ticket/829 ]
09:50 Sewi This would be nice on W7. A Padre quicklaunch button doesn't get the task, the Perl tasks always show up as a new button with two tasks.
10:03 cognominal joined #padre
10:05 Hyppolit svn: r10242 | azawawi++ | http://padre.perlide.org/trac/changeset/10242
10:05 Hyppolit Added Win32::TieRegistry as a win32 dependency
10:05 Hyppolit trunk/Padre/
10:21 azawawi svn hook seems to be failing
10:21 Sewi Is is possible to load a module which should be started using perl -d: at run time?
10:24 azawawi Sewi: like what?
10:24 azawawi szabgab: pingo...
10:24 azawawi When i commit things now i get the following (after being committed)...
10:24 azawawi Failed to start '/var/subversion/repos/padre/hooks/post-commit' hook
10:29 azawawi where is Gabor btw? is he traveling?
10:37 Sewi yes
10:39 Sewi Austria and fosdom
10:40 Sewi I'd like adding Devel::Trace at Padre run time.
10:40 submersible azawawi:  the post commit hook worked 40mins ago ?
10:47 Alias__ Sewi: You can't runtime load debuggers, or they fuck up any other debugger
10:47 Sewi thx
10:47 Alias__ Sewi: You CAN, however, take a look at Padre::Plugin::BadCode
10:48 Alias__ Which is my new shiny Aspect-Oriented tracing tool (to eventually detect bad usages of code at runtime)
10:48 Alias__ submersible: ... and is currently acting as a dry-run for work :)
10:48 azawawi submersible: yup... it stopped now
10:48 Sewi Alias: You added the () brackets to "sub PADRE_REVISION () { ". Critic is complaining about this and I'd like to understand why you added them.
10:51 kaare joined #padre
11:13 azawawi Alias: ping
11:13 azawawi Alias: could you explain to me what was the problem with padre.exe and the required restart?
11:30 Alias__ azawawi, the root cause is that when you start with padre.exe, you don't have any kind of STDOUT and STDERR
11:31 Alias__ So anything at all like IPC::Run3 etc that temporarily hijack STDOUT and STDERR blow up because every single one expects STDOUT and STDERR to exist
11:31 Alias__ In the case of Padre, that meant that the ORLite::Migrate schema modification scripts would fail to run
11:32 Alias__ Which means Padre would start up expecting a database schema that didn't exist
11:32 Alias__ And from there the world basically exploides
11:32 Alias__ When you ran it as padre.bat, the schema got set up
11:33 azawawi i see
11:33 Alias__ When you switched back to padre.exe, the problem still existed, but because it doesn't have to RUN anything during startup, it would survive at least long enough to put the editor on the screen
11:34 Alias__ So what we need to do, somehow, is have either a real or fake STDOUT/STDERR added during early startup
11:34 azawawi or create a hidden console window...
11:34 Alias__ So that the IPC::Run3 STDOUT/STDERR hijacking works
11:34 Alias__ It doesn't need to be a command line
11:34 Alias__ It just needs to be something real
11:34 Alias__ We could even dump to two temporary files
11:35 Alias__ I think that as long as fileno() works, that's enough
11:36 Alias__ You might even be able to try a fake Tie'd one
11:36 azawawi Sewi: didnt we do a VBS script that hid console windows?
11:37 azawawi Alias: i also figured out what we need to do in the case of a UAC action...
11:37 azawawi Alias: launch it in a new process that is elevated...
11:37 azawawi Alias: and if it is in the UI, we need to add the shield flag to the button...
11:38 azawawi Alias__: i'll see what i can do then
11:39 Alias__ ah
11:39 Alias__ I've almost got the Padre::Startup stuff finished
12:15 LeK joined #padre
12:36 submersible joined #padre
13:10 danlucraft1 joined #padre
13:53 Alias__ OK, I think I've finished the new startup code
14:03 submersible mmm
14:04 Alias__ Well, almost
14:04 Alias__ It just crashed
14:05 Alias__ oooo baby it's fast
14:06 cognominal joined #padre
14:06 jq Alias__: i was asking whether it wouldn't be a good idea to introduce "module" or "cpan" in ordb-packaged-mandriva
14:06 Alias__ What do you mean "module" or "cpan" ?
14:06 jq ordb-module-packaged-mandriva
14:06 Alias__ ah
14:06 jq or ordb-module-mandriva
14:06 jq or ordb-cpan-packaged-mandriva
14:07 jq well, you see
14:07 jq wdyt?
14:07 Alias__ ordb-cpan-mandriva would be better
14:07 Alias__ imo...
14:07 jq ordb-cpan-mandriva then
14:07 Alias__ Double-namespacing is usually not needed
14:07 jq i don't care, but i want sthg which is understandable
14:07 Alias__ Foo::X::Y::Bar
14:07 jq and ordb-packaged-mandriva was a bit confusing
14:07 Alias__ Unless you plan to see Foo::Z::Something
14:08 Alias__ AND Foo::X::Something
14:08 Alias__ AND Foo::X::Y::Something
14:08 Alias__ Then the use of two namespaces is excessive
14:08 jq well, it depends whether you already use ordb-cpan in all your modules :-)
14:09 Alias__ Even then, ordb-cpan-rt sits fairly naturally alongside of ordb-cpan-mandriva
14:09 jq it will be ORDB-CPAN-Mandriva, then
14:10 Alias__ Sounds good
14:44 danlucraft joined #padre
14:50 Alias__ Hey, no message
14:50 Alias__ Oh right, the server
14:50 Alias__ The new big startup feature is landed
14:54 isec joined #padre
14:56 LeK Alias, after looking at the problem, it looks like problems with"alarm" and "system" calls with strawberry on win7... is this an uncommon configuration for padre?
15:05 ilbot2 joined #padre
15:05 Topic for #padreis now Latest Padre is 0.55, for installation see http://padre.perlide.org/trac/wiki/Download | Log: http://perlide.org/irclogs/ and http://irclog.perlgeek.de/padre/today | Pasting: http://scsys.co.uk:8001
15:05 CSJewell2 joined #padre
15:05 Alias_ joined #padre
15:05 jayk__ joined #padre
15:05 kthakore_ joined #padre
15:05 LeK_ joined #padre
15:05 SvenDowideit__ joined #padre
15:05 jjore joined #padre
15:05 shadowpaste joined #padre
15:06 mj41 joined #padre
15:08 PerlJam joined #padre
15:08 Hinrik joined #padre
15:08 jq joined #padre
15:09 asphere joined #padre
15:17 cognominal joined #padre
15:19 awnstudio joined #padre
15:22 basti1985 joined #padre
15:23 basti1985 left #padre
15:33 jq Alias_: ordb-cpan-mandriva is released
15:33 jq with an up-to-date database
15:33 jq available on the net
15:33 Alias_ excellent
15:33 jq now it's up to you to do what you want
15:33 jq i've done my part :-)
15:34 Alias_ Sure thing
15:35 jq i need to schedule db build + update on the net on a regular basis
15:35 Alias_ Hopefully I can build a "Mandriva Most Wanted" top 100 list
15:36 jq that would be nice
15:44 jq Alias_: btw, any "problem" or remarks would be nice
15:47 Alias_ ok
15:47 Alias_ Once I've tried it
15:50 jq both on the db (form, pb, etc.) and the data itself (obvious pkg missing, etc.)
15:50 LeK Alias, on Debug::Client, I'm doing step by step digging, and I wonder if in t/01-add.t, the "my $debugger = start_debugger();" line should be blocking ? (line 16)
15:52 Alias_ Runs in under a second for me
15:52 LeK (forget it, it doesn't blocking when called from the good working directory)
15:52 * Alias_ bed
15:53 Corion joined #padre
15:53 Corion Hi all!
15:54 LeK hi
15:54 Corion Ctrl+Tab behaviour is still broken, at least on Win32 - I'm about to introduce an on_ctrl_tab handler which then properly dispatches, as the Wx shortcuts in Padre::Action::Window interfere with the Ctrl-Tab behaviour switching un Padre::Wx::Main
15:55 Corion But I'm not sure whether yet another layer of indirection is what we want. IMO, the whole dual behaviour of Ctrl-Tab should go, as Ctrl-Tab should map to window.next_file and window.prev_file as it does in all other UIs.
16:08 Sewi joined #padre
16:08 Corion Hi Sewi ;)
16:08 Sewi Hi
16:08 Corion I was just ranting about how Ctrl-Tab is still broken, at least for Win32, because the ::Action::Window shortcuts interfere with the switching in the on-key handler
16:08 Sewi I completed your ticket (which you didn't create) :-)
16:08 Corion Yeah, I saw that - cool :)
16:09 Corion I'd like to either rip out the non-standard behaviour or introduce on_ctrl_tab, which acts as a central switch
16:10 Corion on_ctrl_tab would be yet another layer of indirection, while ripping out the "prev pane" behaviour loses a feature. That feature also exists on ctrl-shift-p or something I think.
16:10 Corion I think you introduced the new behaviour so I think you should get a word in before I rip the feature out :-))
16:10 LeK I traced down the problem of the first test (and probably all other tests) of Debug::Client... it seems related to an error reported in Term::ReadLine when debugger starts...
16:13 Corion Shouldn't we set an environment variable so the debugger doesn't even try to load Term::ReadLine?
16:14 Sewi Ctrl-Shift-P is scheduled for removal - because Ctrl-Tab is there
16:14 LeK this readline.pm is rather heavy, the error found is in a line about SetTTY  ## Put Into raw mode
16:16 Corion LeK: Yes - I think you can set $ENV{PERL_RL} to something that makes Readline behave tame. I think. I remember fighting this when cpantesters ran tests for my shell :)
16:16 Corion I'll look at what I did then, to make Term::ReadLine not fail just because there is no terminal
16:17 Corion Sewi: So you're in favour of adding yet another layer of indirection to centralize the distinction between window.prev_pane / .next_pane and .last_used_pane?
16:17 LeK that would be a good idea... at least under win32... it's strange i'm like the only one having the problem
16:18 Sewi The Ctrl-Tab action has a shortcut definition (like every action has). I won't do a on_ctrl_tab but fire the right action of the on-key-handler gets something.
16:18 Sewi We're working towards configurable key bindings - and on_ctrl_tab breaks with this.
16:19 Corion Sewi: Yes, but that logic already needs to exist in two places, and currently only exists in on_key() , and the other, Wx-automatic binding interferes horribly with that
16:20 LeK set TERM=dumb seems to get rid of the perdebugger warning
16:20 Corion But yes, if you're going to do configurable key bindings everything will need to reflect through on_key anyway, or at least the actions need to reconfigure themselves
16:20 Corion LeK: That sounds sensible to me
16:20 Corion Sorry - gotta run! Have a nice evening all
16:55 LeK it must be a stupid question.... but how do you enable padre's log? i'm stuck on the splashscren and can't find any clue of what's going wrong...
16:57 Sewi Are you using the released version or trunk?
16:57 LeK cpan installed
16:57 LeK 0.55 i think
16:58 LeK (confirmed 0.55 cpan)
17:00 Sewi dev.pl, the trunk start script could do it
17:01 danlucraft1 joined #padre
17:01 Sewi Set env var PADRE_DEBUG=1
17:07 danlucraft joined #padre
17:07 LeK the log is created but empty with the cpan version... let's see the trunk if i have better luck
17:21 Sewi svn?
17:22 Sewi Oh, Hyppolit left :-(
17:27 LeK i can't be able to get it started....
17:33 Sewi What exactly happens?
17:34 LeK exactly... nothing... in fact, i've tried so many ways to get it running that i begin to lose myself
17:34 Sewi Which OS?
17:35 LeK Win7, with the cpan version, I see  the splashscreen and it hangs there
17:36 LeK launching it with perl dev.pl i don't see the splashscreen, but the command hangs for nearly a minute before exiting without any information
17:39 Sewi Try dev.pl --die
17:40 LeK nothing seems to happen for now, the command hangs for 60 seconds and dies
17:40 LeK I suspect some kind of network problem (socket timeout?)
17:41 Sewi Strage
17:41 Sewi Try dev.pl --help
17:41 LeK trying to procmon it, I just see it dealing with winsock registry, open a socket, then nothing
17:41 LeK it's working correctly, I have both series of help hints
17:41 LeK (dev.pl and padre)
17:42 Sewi the debugging and tracing options might help you
17:42 LeK if it was only working
17:42 LeK all I see in the log (with the cpan version) is DEBUG ENABLED with the date... then nothing
17:43 Sewi If nothing else helps, try the full trace
17:43 LeK how should I do this?
17:44 Sewi First try dev.pl --die -t
17:44 Sewi If this won't get anything useful, try dev.pl --die --fulltrace
17:45 LeK command hanging, i suppose it will stop after the 60sec timeout, then i'll try fulltrace
17:45 Sewi No output?
17:45 LeK no output, just hanging ( with --die -t)... i'll try fulltrace right now
17:46 LeK cpan module missing, just wait a second
17:47 LeK now I see a lot of trace... let's wait a second
17:47 Sewi What's last?
17:48 LeK it is running
17:48 LeK splashscreen disappeared, but trace is continuing... actually in action queue
17:49 LeK this time... padre finally appears!
17:49 Sewi lol
17:49 LeK grrrr!
17:49 Sewi Really strange, sounds like a timing problem.
17:49 LeK i'll do some more try
17:49 Sewi Tracing slows down Padre.
17:50 LeK yes, i can see this... let's see another try without trace
17:50 LeK arrrggggg, it's working now, window appears in less than 10 seconds
17:51 LeK this time it seems it hangs again
17:52 LeK maybe I've put the finger on some weird problem...
17:52 Sewi Did it run anytime before?
17:54 Sewi If not, start it with fulltrace, go to Help->About and get the config dir from the Info panel.
17:54 LeK in fact, yesterday night I tried it and it refuses to run, hanging like it is doing right now... then, without any special work, it was working (at least a little, because there were some weird things, like the problem with the debugger)
17:54 Sewi Remove the config.yml and  config.db
17:55 Sewi If you didn't customize any configuration preference, you shouldn't loose anything
17:55 LeK today, i tried another batch of tweaks (with different versions of strawberryperl, etc...), and only met problems... until the two tries that worked a minute ago
17:56 LeK I already removed the config files, but I don't mind to do it again (i'll try to check the about box for getting the right config dir)
17:57 LeK it seems to have hanged with the fulltrace..... somewhere in lib/IO/Handle of strawberry perl
17:57 LeK and it was just after a socket operation... i think we're getting close :)
17:59 LeK It seems that I can reproduce the problem nearly every time
18:01 LeK do you want a pastbin of the trace? should I open a new ticket in the track? (I wanted to report the problem with ReadLine too)
18:02 LeK left #padre
18:02 LeK joined #padre
18:08 Sewi Do you have a trac account?
18:08 LeK nop, according to the wiki, new accounts should be asked for there
18:09 Sewi If we were a commercial support line, I'd point you to the wiki now :-)
18:09 Sewi Registration is open now
18:09 LeK you can still do this... and ask me to fill it
18:10 cognominal joined #padre
18:10 cognominal joined #padre
18:11 Sewi Please try to create a ticket and you should be able to sign up now.
18:11 LeK ok, i registered
18:11 Sewi thanks
18:12 Sewi registration closed
18:12 root joined #padre
18:12 LeK you want a bug report both for the hanghup and the Debug::Client test and readline problems?
18:13 Sewi ping Alias, garu, Ryan52: Would you give my second personality (sewi_) op, please? It's logged in from a server and should be able to keep it until szabgab returns and restarts Hyppolit
18:17 Sewi Lek: feel free
18:23 LeK bug report done for the hangup
18:23 LeK Ticket #830
18:27 Sewi Oh, another idea: Did you try running Padre as Administrator?
18:32 LeK i tried without success, but not with the trunk version, i even tried to add perl &co in the firewall, with firewall disabled, etc...
18:32 LeK let's see with trunk version
18:34 LeK not any better
18:36 LeK arrrggggggggggg!!!!!!!!! netbeans was listening on port 4444!
18:37 LeK after closing netbeans, padre alwsays starts
18:52 Sewi Add it to the ticket, please.
18:52 Sewi Others might have the same problem
18:58 LeK yep, I changed the criticity of the problem and added the information on the diagnostics... maybe better error detection could be added
18:59 LeK anyway, thank you very much for your help
19:07 Ryan52 Sewi: there you go
19:44 Sewi Thanks, Ryan52
19:45 Sewi I'd prefer having Hyppolit back but we might need to wait until szab returns
20:01 robn joined #padre
21:13 SvenDowideit_ joined #padre
21:14 SvenDowideit joined #padre
23:18 patspam joined #padre
23:47 LeK left #padre

| Channels | #padre index | Today | | Search | Google Search | Plain-Text | summary